Technical Evaluation & Sample Request Guide for Series & Parallel Configuration

What to Specify Before You Send a Single Inquiry #

Most evaluation failures happen before the sample ships. An underspecified inquiry produces a sample that satisfies the factory’s default parameters, not yours. Before contacting any Shenzhen-based pack house or cell supplier, your inquiry document needs to nail down six things: nominal configuration (series count, parallel count, topology), target capacity in Ah at the 0.5C discharge rate (not peak), operating voltage window, maximum continuous charge and discharge current, expected cycle life target with the test rate, and the thermal environment the pack will see in actual use.

That last point gets skipped more often than it should. A 4S3P LFP pack evaluated at 25°C will show meaningfully different impedance growth curves than the same pack cycling at 40°C in a vented enclosure — which is what many industrial portable applications actually see. If your inquiry doesn’t specify this, the factory will send you data at 25°C and call it a day.

For series-parallel configurations specifically, the impedance matching requirement deserves its own line item. State your acceptable cell-to-cell DC internal resistance spread in the inquiry — we typically specify ≤3.5 mΩ spread across the full pack for 4S or higher configurations intended for daily cycling. If the supplier can’t acknowledge that parameter or asks why it matters, that tells you something about their sorting capability.

One field that often gets left blank on inquiry forms: BMS protection threshold alignment. Specify your required cell overvoltage cutoff (typically 3.65V for LFP), undervoltage cutoff, and overcurrent trip point. A factory that builds both consumer and industrial packs may have several BMS firmware variants. Without specifying, you’ll get whatever they shipped last week.

Early in 2024, we rolled out what our team calls the SP-IQ form (Series-Parallel Inquiry Qualification form) — a one-page technical brief that forces engineers to specify all of the above before any supplier contact is initiated. The number of unusable first samples dropped noticeably after that process was standardized.

The Root Cause of Mismatched Evaluation Samples #

The non-obvious failure mode that derails more evaluations than any other: capacity grade mismatch between the sample lot and the cells that will actually ship at production volume.

Here’s the mechanism. Tier-2 and Tier-3 Dongguan cell manufacturers sort their production output into capacity grades — typically A, A-, and B — but the boundaries of those grades shift depending on how full the warehouse is and what large OEM orders are pending. When you request 20 evaluation cells, you’re a low-priority customer pulling from available stock. The factory’s picker will select cells from whatever bin is accessible. If the A-grade bin is depleted by a large OEM order that week, you may receive A- cells that pass their internal spec (say, ≥95% of nominal capacity) but won’t represent the actual grade you’d receive in a 5,000-unit production order — which might come from a different production batch entirely.

This matters acutely for series-parallel configurations because series strings amplify capacity mismatch. In a 4S pack, if even one cell in the string is running 3.1% below the others, that cell will hit its undervoltage cutoff first on every discharge cycle. The BMS cuts the entire string. Usable capacity drops, cycle count appears to degrade faster than the cell chemistry warrants, and the engineering team may incorrectly conclude there’s a BMS tuning problem when the root cause is procurement-side cell sorting.

The measurement method to confirm grade consistency: discharge each sample cell individually at 0.5C from full charge (per your specified upper voltage) to the lower cutoff, measure delivered Ah, and compare cell-to-cell. In a properly graded A-grade LFP lot (280Ah prismatic, EVE or CATL-tier), the spread should be ≤1.4Ah across a sample of 10 cells. If you’re seeing ≥2.1Ah spread in a 10-cell sample, the lot is mixed-grade regardless of what the test report says. We’ve seen this spread on samples labeled “Grade A” from three separate suppliers in the same calendar quarter — two of them had IEC 62619-compliant documentation otherwise perfectly in order.

The IEC 62619:2022 secondary lithium cells and batteries standard addresses safety requirements but does not mandate cell-to-cell capacity sorting tolerances. That gap is one you have to close contractually.

Corrective Actions Ranked by Impact and Feasibility #

When your evaluation sample reveals problems — capacity spread, impedance mismatch, BMS protection thresholds misaligned with your spec — there are five paths forward, and they are not equally useful:

1. Request a second sample lot with explicit sorting criteria attached. This is the fastest corrective action and resolves roughly 60% of capacity mismatch issues in our experience. Write the re-request specifying: individual cell capacity ≥98% of nominal, DC internal resistance ≤X mΩ per cell (specify your number), and request the cell-level test report for each unit in the lot. This costs you two to three weeks of elapsed time, no tooling cost. If the supplier can’t provide cell-level test data, move to option 4.

2. Require witnessed outgoing QC at the factory. Send your own engineer or a third-party inspection agent to witness the cell sorting and pack assembly for the evaluation lot. This adds cost (inspection fees in Shenzhen run approximately ¥1,800–2,400/day for a qualified battery inspector) but eliminates the sample-quality lottery. This approach works best once you’ve confirmed a supplier is technically capable but has inconsistent process discipline.

3. Perform full impedance spectroscopy incoming. If you have EIS (electrochemical impedance spectroscopy) capability in-house, screen every incoming cell at 1 kHz AC impedance before assembly. The threshold for rejection in a 4S configuration: any cell with impedance more than 18% above the lot median. This gives you a quantitative gate that doesn’t rely on the supplier’s paperwork at all. Capital cost for a basic impedance tester: $3,500–6,000. This is worth the investment if you’re running more than 15 evaluation programs per year.

4. Switch to a supplier with documented cell sorting SOP. Ask for their SP-level grading procedure (or equivalent internal document) as a qualifying step. Suppliers with genuine A-grade sorting capability will have it written down and will share it. Those without it will give you a verbal assurance. Verbal assurances on cell grading are worth nothing for series-parallel configurations where a single outlier cell degrades the whole pack.

5. Redesign the pack topology to add parallel redundancy. Adding a parallel leg (e.g., going from 4S1P to 4S2P) reduces the capacity sensitivity of any single underperforming cell. This fixes symptom, not cause, and requires more cells per pack, but it’s a legitimate engineering trade-off for applications where supplier diversity is constrained. For battery pack design decisions around series vs. parallel topology, the trade-off analysis runs deeper than just cell count.

What to Specify Upfront to Prevent Grade Mismatch at Scale #

Put these directly in your PO and supplier brief, not in a side email:

• Minimum individual cell capacity: 98% of nominal Ah, verified at 0.5C/25°C per UL 1973 Section 8 capacity test method
• Maximum cell-to-cell DC IR spread within a single batch: 3.5 mΩ
• Batch certificate required: cell-level test data for each production run code, not a blanket lot certificate
• BMS firmware version must be documented and version-locked for your SKU
• Cycle life verification: 500 cycles at 1C/1C, 80% capacity retention minimum, per IEC 61960-3 test conditions

The document to request before placing any production order: the cell-level incoming inspection report from the factory’s own goods-receiving QC, showing individual capacity and IR data for the specific cells assigned to your production run code. If that document doesn’t exist, the factory is not running a sorted assembly process.

Evaluation Timeline: Inquiry to Design-In Decision #

A realistic timeline for evaluating a new series-parallel configuration supplier, assuming no major surprises:

Total: 10–15 weeks from first inquiry to signed supply agreement. Any factory that promises a design-in timeline under 8 weeks is either skipping cycle testing or working from pre-existing data on a related SKU. Acceptable in some cases — but confirm which.

The abbreviated 50-cycle test is a triage tool, not a qualification. For safety-relevant BMS engineering decisions and protection threshold validation, the 50-cycle curve gives you early-stage impedance growth data that’s predictive but not conclusive. For mission-critical applications, extend to 200 cycles minimum before committing to a production supply agreement.

One industry observation that’s shaping evaluation timelines in 2025: the lead time for Grade-A 280Ah LFP prismatic cells (EVE/CATL-equivalent specification) from Shenzhen integrators has compressed back to 3–4 weeks from the 8–12 week delays seen in 2022–2023. That means the evaluation cycle is now the long pole, not cell availability — which actually creates pressure on engineering teams to shortcut the cycle testing phase. Resist that pressure. The UN 38.3 transport test requirements cover transport safety, not field reliability. Those two things are not the same.

Sourcing Guidance for Buyers #

When evaluating Shenzhen-based pack suppliers for series-parallel configurations, the first document to request is the factory’s cell-level incoming QC data for a recent production run — not the cell manufacturer’s datasheet, and not a blanket COA. A supplier that tracks cell-level IR and capacity at goods-receiving has a real process. One that offers only a supplier-issued COA is passing along the cell maker’s claim without independent verification.

The qualification red flag specific to this category: a factory that can’t tell you which cell bins were used for your evaluation sample. If they can’t trace sample cells back to a specific production lot with individual test data, the sample is not representative of anything.

For incoming inspection of production samples, pull a minimum sample of 12 cells per 500-unit batch. Test each cell individually for capacity at 0.5C discharge and DC internal resistance at full charge. Reject the batch if any cell falls below 97% of nominal capacity or if cell-to-cell IR spread exceeds 4.2 mΩ. These thresholds are tighter than most factories will propose — that’s intentional. Pack performance in series configurations is disproportionately limited by the weakest cell, so your incoming gate needs to filter harder than the factory’s outgoing gate.

---

Published by compactbess.com Technical Team | Request a sourcing consultation